The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied UI and layout descriptions to generate HTML files, which represents an indirect prompt injection surface. This is effectively mitigated by the skill's foundation rules, which require the use of textContent and createElement for DOM construction and explicitly forbid innerHTML for variable data. Additionally, the skill specifies that submissions should be treated strictly as data and never interpreted as new instructions.
[SAFE]: The skill uses a local loopback (localhost) for data submission in local environments, ensuring that generated content and user choices remain within the local context. It also includes a fallback to clipboard-based submission if the local server is unavailable.
[SAFE]: No remote code execution patterns, obfuscation, or unauthorized network operations were identified. External resource usage is limited to well-known services like Google Fonts, which are considered safe.

html-brainstorm-grid